Re: [monit] Issue monitoring winbindd redux

[Martin Pala]

Hi,

it seems that the service started, but was not accepting the  
connections yet, whereas monit tried to perform the tcp check  
immediately after the process started. I recommend to try the  
monit-5.0 (http://www.tildeslash.com/monit/dist/beta/) since there are  
two changes which may help.

If the problem will persist, please send the strace output for monit  
trying to start the winbindd:

strace -f -p <monit's pid> > 2>&1 | tee monit.strace


Thanks,
Martin


On Aug 14, 2008, at 6:10 PM, Dimitri Yioulos wrote:

> All,
>
> I originally posted this early last week, but am not sure if it got
> distributed.  Anyway, it's still an unresolveed issue, and I'd  
> appreciate
> your help with it:
>
> I've installed monit on a CentOS 5.2 box to keep an eye on winbind (a
> component of samba), which is used in conjunction with our VPN for
> authentication purposes.  Note that I am not running the smb  
> component of
> samba, as I'm not looking to do any file sharing.  This scheme works  
> fine;
> VPN authentication has worked fine for the past three years.  Of  
> course, my
> purpose for using monit is to make sure that the winbind daemon, and  
> thus the
> VPN, is always available.
>
> Here's the winbind directive in monirc or monit.conf:
>
> check process winbindd with pidfile /var/run/samba/winbindd.pid
>    group samba
>    start program = "/etc/rc.d/init.d/winbind start"
>    stop  program = "/etc/rc.d/init.d/winbind stop"
>    if failed host 192.168.1.5 port 139 type TCP then restart
>    if 5 restarts within 5 cycles then timeout
>
> Pretty straighforward.
>
> If I stop winbind, this is what monit reports in monit.log:
>
> [EDT Aug  4 10:01:32] error    : 'winbindd' failed, cannot open a  
> connection
> to INET[192.168.1.5:139] via TCP
> [EDT Aug  4 10:01:37] info     : 'winbindd' trying to restart
> [EDT Aug  4 10:01:37] info     : 'winbindd' stop: /etc/rc.d/init.d/ 
> winbind
> [EDT Aug  4 10:01:38] info     : 'winbindd' start: /etc/rc.d/init.d/ 
> winbind
> [EDT Aug  4 10:03:38] info     : Monit has not changed
> [EDT Aug  4 10:03:38] error    : 'winbindd' failed, cannot open a  
> connection
> to INET[192.168.1.5:139] via TCP
> [EDT Aug  4 10:03:38] info     : 'winbindd' trying to restart
> [EDT Aug  4 10:03:38] info     : 'winbindd' stop: /etc/rc.d/init.d/ 
> winbind
> [EDT Aug  4 10:03:39] info     : 'winbindd' start: /etc/rc.d/init.d/ 
> winbind
> [EDT Aug  4 10:05:39] error    : 'winbindd' failed, cannot open a  
> connection
> to INET[192.168.1.5:139] via TCP
> [EDT Aug  4 10:05:39] info     : 'winbindd' trying to restart
> [EDT Aug  4 10:05:39] info     : 'winbindd' stop: /etc/rc.d/init.d/ 
> winbind
> [EDT Aug  4 10:05:40] info     : 'winbindd' start: /etc/rc.d/init.d/ 
> winbind
> [EDT Aug  4 10:07:40] error    : 'winbindd' failed, cannot open a  
> connection
> to INET[192.168.1.5:139] via TCP
> [EDT Aug  4 10:07:40] info     : 'winbindd' trying to restart
> [EDT Aug  4 10:07:40] info     : 'winbindd' stop: /etc/rc.d/init.d/ 
> winbind
> [EDT Aug  4 10:07:41] info     : 'winbindd' start: /etc/rc.d/init.d/ 
> winbind
> [EDT Aug  4 10:09:41] error    : 'winbindd' service timed out and  
> will not be
> checked anymore
>
> I first tried monit-4.10-1, then reverted to monit-4.9-1, and both  
> give the
> same result.  On other boxes (centOS 3.9 and CentOS 4.6), I have monit
> watching both smb and winbind (same directive for winbind) without  
> issue.
>
> I've searched high and low for the solution, but can't find one.   
> Can anyone
> help?
>
> Thanks.
>
> Dimitri
>
> --
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
>
>
>
> --
> To unsubscribe:
> http://lists.nongnu.org/mailman/listinfo/monit-general